[OPSAWG] [IANA #1240167] IANA question regarding draft-ietf-opsawg-ipfix-srv6-srh-01

[Amanda Baber via RT <[e-mail]>]

Hi Benoit,

I took this to our technical director, James Mitchell, and he wrote, "These segment URLs [ipfix.xhtml#something] are not great for consumption by automated systems ... Segments are managed by the client, therefore the 'collector' would need to extract the segment from the URL, request the resource from the server, parse it, then locate the element based on the segment. Do-able, but not great.

"I'd always like to understand what they're after, and how we can find a solution with what we currently provide. They may be able to use the XML source files to achieve what they want; alternatively the CSV files are per-registry."

He won't be in London, but could join a call if you want to meet there. Alternatively, if you'd like to schedule a call earlier, just let me know ([e-mail] if you want to take this outside the ticketing system). More below.

On Thu Oct 06 15:31:00 2022, [e-mail] wrote:
> Hi Amanda, IPFIX IE doctors,
> 
> See inline.
> 
> On 9/30/2022 4:58 AM, Amanda Baber via RT wrote:
> > Hi Benoit, all,
> >
> >> Dear IPFIX doctors, (IANA),
> >>
> >> We would like to get your feedback regarding the IANA section in
> >> draft-ietf-opsawg-ipfix-srv6-srh-01.
> >> Especially, the two following information elements:
> >> srhFlagsIPv6:
> >> https://datatracker.ietf.org/doc/html/draft-ietf-opsawg-ipfix-srv6-
> >> srh-01#section-5.1
> >> srhSegmentEndpointBehavior:
> >> https://datatracker.ietf.org/doc/html/draft-ietf-opsawg-ipfix-srv6-
> >> srh-01#section-5.11
> > We can keep separate registries in sync (although we don't currently
> > have automation to ensure this), but is the intention for the IPFIX
> > IPv6 SRH Flags and IPFIX SRV6 Endpoint Behavior registries to be
> > contained within each IPFIX IE registration's Description field?
> >
> > In 2020, with IE Doctor approval, all IPFIX IE Description field
> > tables that constituted sub-registries were replaced with links to
> > separate sub-registries located outside of the IPFIX Information
> > Elements registry. You can see the list of sub-registries under the
> > heading "Registries included below":
> >
> > https://www.iana.org/assignments/ipfix
> >
> >> I went to the IANA table in Philly and we discussed those. Hence I
> >> copied IANA here.
> >> In the currentdraft-ietf-opsawg-ipfix-srv6-srh-01
> >>  <https://datatracker.ietf.org/doc/html/draft-ietf-opsawg-ipfix-
> >> srv6-
> >> srh-01>
> >> version, we created two IPFIX subregistries, which mimic existing
> >> Segment Routing registries.
> >> The main reason is that we are in favor of having a self contained
> >> IPFIX
> >> IANA registry, which we can download as a cron job in the collector.
> >> We
> >> discussed such a project with Michelle Cotton in the past (I know
> >> that
> >> Michelle moved on).
> > I'm afraid we don't have any of Michelle's notes on this topic. What
> > will you need IANA to do? We may need to put you in touch with IANA's
> > technical director. In the future, the registries will be moved from
> > an XML-based to a database-based registry system.
> Regarding the argument of "having a self having a self contained IPFIX
> IANA registry, which we can download as a cron job in the collector",
> what we need is the ability to retrieve either the entire registry in
> one go, or have pointers to other (registries) the IPFIX collector has
> to take into account ... in an automatic manner
> The former, with subregistries duplication (and synchronization) might
> be beyond repair at this point in time (at least, I don't have the
> courage to engage), on top of pushing much administration to IANA for
> the synchronization/maintenance
> 
> Looking closer at the IFPIX registry, let me provide a logic that
> might
> work, without too much changes IMO.
> 
> Let's say I take care of a IPFIX collector, which I need to update on
> regular basis, there are different cases:
> 1. A new IPFIX IE is added in the registry
>     Ex: Let's say I specify in IANA the IPFIX IE 492 = "my-new-field"
>      I download the entire registry as a cronjob, install it in my
> collector, and I can now understand all flow records that contain the
> new IPFIX IE 492
> 2. A new value is added for one of the IPFIX sub-registries
>      Ex: Let's say I add value 25 = "my-new-classification-engine-id"
> at
> https://www.iana.org/assignments/ipfix/ipfix.xhtml#classification-
> engine-ids
>     What is the logic?
>      I download the entire registry as a cronjob, parse the file, when
> I
> arrive at IPFIX field 101 (classificationEngineId) => I do a lookup on
> *[https://www.iana.org/assignments/ipfix/ipfix.xhtml#classification-
> engine-ids]
> in the Description field* , read the new value 25, install it in my
> collector, and I can now understand all flow records that contain a
> classificationEngineId value 25.
> 3. A new value is added for in the external (non-IPFIX) registry
>      Ex: a new port is added to
> https://www.iana.org/assignments/service-names-port-numbers].
>      The following IPFIX IEs refer to this registry:
> sourceTransportPort, destinationTransportPort, udpSourcePort,
> udpDestinationPort, udpDestinationPort, etc.
>     What is the logic?
>      I download the entire registry as a cronjob, parse the file, when
> I
> arrive at any of those IPFIX IEs => I do a lookup**on
> https://www.iana.org/assignments/service-names-port-numbers*in the
> Addition Information field*, read the new port value, install it in my
> collector, and I can now understand all flow records that contain the
> new port value
> 
> At least the logic could work (even if looking up two fields is not
> ideal from a logic point of view)
> I checked all http links in the IPFIX registry and found 3
> "exceptions"
> - [http://standards.ieee.org/regauth/ethertype/eth.txt], will not
> work,
> but we might consider this one as an exception.
> - portRangeStart and portRangeEnd point to
> https://www.iana.org/assignments/service-names-port-numbers. Not to
> sure
> how to treat this one in an automatic way
> - [https://www.iana.org/assignments/ianaiftype-mib] should be treated
> differently, but that's fine.
> 
> One conclusion is that there are different treatments whether the
> links
> are in the *Description field* or the *Additional Information* field.
> Interestingly, RFC 7102 doesn't mention this *Additional Information*
> field :-) And this is precisely this one that I would like to use
> below,
> since we speak about non-IPFIX registries in
> draft-ietf-opsawg-ipfix-srv6-srh.
> If we don't pay attention to that detail, here is a proposal
> (combining
> Med proposal with the logic above)
> 
> 5.1
>  <https://datatracker.ietf.org/doc/html/draft-ietf-opsawg-ipfix-srv6-
> srh-01#section-5.1>.
> srhFlagsIPv6
> 
> Name:  srhFlagsIPv6
> 
> ElementID:  TBD1
> 
> Description:  This IE identifies the 8-bit flags defined in the SRH.
>        Assigned flags and their meanings are provided in the "Segment
> Routing Header Flags" registry.
> 
> Abstract Data Type:  unsigned8
> 
> Data Type Semantics:  flags
> 
> Reference:  [RFC-to-be],RFC8754
> <https://datatracker.ietf.org/doc/html/rfc8754>
> 
> *Additional reference: ****https://www.iana.org/assignments/ipv6-
> parameters/ipv6-parameters.xhtml#segment-routing-header-flags*
> 
> 
> Note: and similarly for5.9
> <https://datatracker.ietf.org/doc/html/draft-ietf-opsawg-ipfix-srv6-
> srh-01#section-5.9>.
> srhActiveSegmentIPv6Type Note: as you can see, I have a different
> opinion that Med on this one => we need to point to the specific
> registry within the group, for automatic treatment
> 
> Would that work, or IANA/IPFIX doctors will frown upon because
> the*"additional information"*  field is not mentioned in RFC7102?

This is fine for IANA. 

A few years ago, with IE Doctor approval, the fields called "Reference" and "Requester" in RFC 5102 were renamed "Additional Information" and "Reference" (the term used in other registries for RFCs and people who request registrations). 

thanks,
Amanda 

> Regards, Benoit
> 
> >
> >> On top of that, it might be beneficial for the
> >> IPFIX
> >> experts to review any changes coming from a registry we mimic, just
> >> to
> >> see if there are no new inconsistencies from an IPFIX point of view.
> >>
> >> However, Med, in cc., has a valid point that the current IPFIX IANA
> >> entries are inconsistencies already.
> >> Ex: destinationTransportPort points to a different registry
> >> Ex: tcpControlBit. See
> >> https://datatracker.ietf.org/doc/draft-boucadair-opsawg-rfc7125-
> >> update/
> >> So he advocated that we don't duplicate, with an IPFIX subregistry,
> >> information stemming from a different source. Pointing to the
> >> existing
> >> registry (ex: "Segment Routing Header Flags") + a RFC reference is
> >> sufficient for him. Solving the self-contained IPFIX registry issue
> >> would be a (too) huge job at this point in time.
> >>
> >> This is what we currently have in the draft:
> >>
> >> 5.1
> >>  <https://datatracker.ietf.org/doc/html/draft-ietf-opsawg-ipfix-
> >> srv6-
> >> srh-01#section-5.1>.
> >> srhFlagsIPv6
> >>
> >> Name:  srhFlagsIPv6
> >>
> >> ElementID:  TBD1
> >>
> >> Description:  This Information Element identifies the 8-bit flags
> >>     defined in the SRH.  Values for this Information Element are
> >>     listed in the "IPFIX IPv6 SRH Flags" registry, see [IANA-IPFIX
> >>  <https://datatracker.ietf.org/doc/html/draft-ietf-opsawg-ipfix-
> >> srv6-
> >> srh-01#ref-IANA-IPFIX>].
> >>     srhFlagsIPv6 values must not be directly added to this "IPFIX
> >> IPv6
> >>     SRH Flags" registry.  They must instead be added to the "Segment
> >>     Routing Header Flags" registry.  Both the "IPFIX IPv6 SRH Flags"
> >>     and the "Segment Routing Header Flags" registries must be kept
> >> in
> >>     synch.  Initial values in the registry are defined by the table
> >>     below.
> >>
> >> +--------+-------------------+--------------------------------------
> >> +
> >> | Value  |    Description    |              Reference
> >> | |
> >> +--------+-------------------+--------------------------------------
> >> +
> >> | 0-1    | Unassigned        |
> >> | |
> >> +--------+-------------------+--------------------------------------
> >> +
> >> | 2      | O-flag            |  [RFC-ietf-6man-spring-srv6-oam-13]
> >> | |
> >> +--------+-------------------+--------------------------------------
> >> +
> >> | 3-7    | Unassigned        |
> >> | |
> >> +--------+-------------------+--------------------------------------
> >> +
> >>
> >> Table 2: "IPFIX IPv6 SRH Flags" registry
> >>
> >>
> >> Note to IANA:  Add a note to the "IPFIX SRV6 Endpoint Behavior"
> >>     registry so that new values are echoed in the new "IPFIX SRv6
> >>     EndPoint Behavior
> >>
> >> Abstract Data Type:  unsigned8
> >>
> >> Data Type Semantics:  flags
> >>
> >> Reference:  [RFC-to-be],RFC8754
> >> <https://datatracker.ietf.org/doc/html/rfc8754>
> >>
> >> This is what Med proposes:
> >>
> >> 5.1
> >>  <https://datatracker.ietf.org/doc/html/draft-ietf-opsawg-ipfix-
> >> srv6-
> >> srh-01#section-5.1>.
> >> srhFlagsIPv6
> >>
> >> Name:  srhFlagsIPv6
> >>
> >> ElementID:  TBD1
> >>
> >> Description:  This IE identifies the 8-bit flags defined in the SRH.
> >>        Assigned flags and their meanings are provided in the
> >> "Segment
> >> Routing
> >>        Header Flags" registry.
> >>
> >> Abstract Data Type:  unsigned8
> >>
> >> Data Type Semantics:  flags
> >>
> >> Reference:  [RFC-to-be],RFC8754
> >> <https://datatracker.ietf.org/doc/html/rfc8754>
> > When IANA links to this registry, will the link have to point to,
> > e.g.,https://www.iana.org/assignments/segment-routing/segment-
> > routing#the-specific-registry, or would it be sufficient to point
> > tohttps://www.iana.org/assignments/segment-routing  (the registry
> > group, rather than the specific registry within it)?
> >
> >> We basically agree that a registry lookup is required for the IPFIX
> >> Collector.
> >> An IPFIX Exporter will export what he sees, regardless of the
> >> semantic
> >> or an IANA registry. The IPFIX Collector will report a potential
> >> problem
> >> if the observed value is not in the IANA registry (bug, IANA entry
> >> hijacked, another convention => if value not observed, I'll send an
> >> error code instead, etc)
> >>
> >> Bottom line: we have two different ways to model the srhFlagsIPv6
> >> and
> >> srhSegmentEndpointBehavior in IANA, with or without an IPFIX
> >> subregistry.
> >> Can you share your views on the best way to register those IEs.
> >>
> >> Thanks and regards, Benoit
> > thanks,
> >
> > Amanda Baber
> > IANA Operations Manager
> >
> >